This site is intended for health professionals only

Prognostication in palliative care

RCGP clinical champion for end-of-life care Professor Keri Thomas on how best to anticipate a dying patient's care needs

Prognostication is always difficult, especially when faced with the tortured question: ‘How long have I got?'

Although this is often asked or inferred, what's possibly really being expressed is an underlying cry for help.

The RCGP end-of-life care strategy explicitly states that this is part of the core business of general practice, adding: ‘The GP and the primary care team occupy a central role [which] is greatly valued by patients and remains pivotal to the effective provision of all other care.'1

So when estimating prognosis in general terms, we also need to think about what we can do to help ensure all patients live as well as possible until they die, and that they die well in the place and manner of their choosing. I think this is the real issue behind the question, and why we need to be one step ahead by predicting likely trajectories of illness, and anticipating and meeting patient needs to enable best-quality care for this group.

Identifying patients in their last year of life

About 1% of the population die each year, but it is inherently difficult to identify which patients may be in their last year of life. The Department of Health end-of-life care strategy emphasises the importance of trying to recognise these patients in order to co-ordinate care well, using tools such as the Gold Standards Framework (GSF).

New support materials will soon be available on the website,, along with a new patient charter. A key component of the GSF is a set of prognostic indicators which use specific clinical signs, a patient's choices and the clinician's intuition as triggers for changes in care.

The National Primary Care Audit in End-of-Life Care was carried out in 15 PCTs and 502 GP practices, and involved 4,500 patient deaths over two months in 2009.2 Its three key findings were:

• only a quarter of patients with relevant deaths were identified and included on the palliative care register

• only a quarter of patients on the register had diseases other than cancer, despite three-quarters of patients overall dying from conditions other than cancer

• patients on the register received better co-ordinated care towards the end of life, with more proactive planning and anticipatory prescribing, and they were more likely to die where they chose.

If patients were identified earlier and placed on the register, it would enable earlier discussion of their wishes and care could be better aligned to preferences, reducing crises. This would lead to fewer hospitalisations and reduced costs.

Identifying the last days, weeks and months

A key part of end-of-life care is identifying which patients are nearing the final months, weeks and days of life, anticipating the likely trajectory of their illness and planning care accordingly.

There are three main such trajectories, in addition to sudden, unpredictable death:

• rapid – typically cancer

• erratic – typically organ failure

• protracted – typically frailty, co-morbidity and dementia.

A needs-based coding system is recommended – using these likely trajectories – to make it easier to provide proactive support at each relevant stage (see box, attached). By capturing patients' needs and concerns earlier through formalised and recorded advance care planning discussions, there is a greater chance of co-ordinated care, ensuring the right thing happens at the right time, with more patients enabled to die where they choose.3 These discussions also give patients a greater sense of control and greater satisfaction with their care, and often increase – rather than diminish – a sense of hope.4

Useful prognostic indicators

How can we identify these patients earlier? In 2005, the GSF team was asked to produce prognostic indicator guidance to support the QOF palliative care indicators on palliative care registers and regular patient discussions at team meetings.

This guidance is regularly updated and a new version is expected soon. The needs-based coding system described above is used to indicate activities at each stage to make care more co-ordinated.

The three main prognostic triggers are mentioned in the box, left, and there are examples of suggested clinical indicators for individual diseases in the box below it. But it is important to also consider measures of functioning.

This guidance is not designed to enable us to answer the explicit: ‘How long have I got?' But it does allow us to focus care on a patient's needs.

Improving skills

The GSF materials are considered to be best practice and its related training programmes were initially developed in primary care, but are now available for care homes, hospitals and more recently domiciliary care.

Most practices are at the foundation level GSF, which is helpful to improve care as outlined in the QOF. But the new primary care training programme Going for Gold enables practices to move to the next stage of deeper implementation and greater consistency and effectiveness – supported by audit and home-grown practice protocols and processes. For more details go to or email

With increasing emphasis on end-of-life care, the advent of GP commissioning and the need for better use of scarce resources, this is an important time for GPs to improve care for people nearing the end of life.

As the RCGP strategy also states: ‘The importance of the holistic role of the family doctor is poised to come into its own in a way never previously encountered.'

Professor Keri Thomas is national clinical lead for the Gold Standards Framework Centre, honorary professor in end-of-life care at the University of Birmingham and RCGP clinical champion for end-of-life care

The International Society of Advance Care Planning and End of Life Care is holding its 2011 conference in London from 22-24 June, and one of the themes is the primary care community. For further details go to

Elderly patient with a doctor Elderly patient with a doctor Clinical triggers

Patients with co-morbidities or other predictors of end-stage illness
• Significant weight loss (>10% over six months)
• Serum albumin <25g/l
• Reduced performance status
• Patients for whom the ‘surprise question' is applicable.

Patients with cancer
Performance status: if the patient is lying in bed more than 50% of the time, prognosis is estimated to be about three months or less. More exact predictors include the palliative performance scale and the palliative prognostic indicator.

Patients with organ failure
1) Heart failure
At least two of the following indicators:
• New York Heart Association stage 3 or 4 – equivalent to moderate or severe
• patient thought to be in the last year of life
(the surprise question)
• repeated admission with symptoms of heart failure
• difficult physical or psychological symptoms despite optimal tolerated therapy.
2) Motor neurone disease
It is difficult to predict any neurological disease progression, but possible MND triggers are:
• swallowing problems
• marked and rapid decline in physical status
• first episode of aspirational pneumonia
• increased cognitive difficulties
• weight loss
• significant complex symptoms and medical complications
• low vital capacity – below 70% of predicted by spirometry.

A fuller version of this box is available at

Prognostic triggers for changes to care

Three triggers are suggested and any can be used to identify patients:
1) The surprise question: ‘Would you be surprised if this patient were to die in the next six to 12 months?' – an intuitive question integrating co-morbidity, social and other factors. If the answer is no, then think about measures that might improve their quality of life now and in preparation for the dying stage. The surprise question can be applied to years/months/weeks/days and trigger the appropriate actions at each stage.
2) Choice/need: the patient with advanced disease makes a choice for comfort care only, not ‘curative' treatment, or is in special need of supportive/palliative care – such as declining renal transplant.
3) Clinical indicators: specific indicators of advanced disease for each of the three main end-of-life patient groups described in the ‘clinical triggers' box.

Needs-based coding system Needs-based coding system Needs-based coding system

click here to download the 'needs-based coding system' table